things i've made
some useful, some experimental,
all fun to look back on ♡
A mix of software, machine learning, data, and experiments I've built with teams and on my own.
Cal Poly senior capstone with Calling All Kids
Calling All Kids Parent Dashboard
Turning children’s simulated calls into a report a parent can actually understand.
The dashboard brings usage patterns, mood and topic signals, temperament, and parent guidance into one calm view. I focused especially on bringing the different report pipelines together, shaping reporting rules, and making that integration easier to test.
- Node.js
- Express
- Flutter Web
- local CSV snapshots
- Mocha
- OpenAI API
Hack4Impact · Paso Robles Food Co-op
PRFC Connect
A communication and events platform used by a real community food co-op.
Paso Robles Food Co-op uses the platform to organize members, groups, messages, referrals, events, and RSVPs. I worked across security, the first authenticated dashboard, accessibility semantics, testing, and a deceptively small event-invite bug as one of two student tech leads.
- Next.js 16
- TypeScript
- Prisma
- MySQL
- Docker Compose
- Vitest
- Playwright
Solo CSC 466 data-mining project
Student Success Prediction
What predicts student dropout before and after college begins?
I built two views of the same question: one using only information available around admission, and one adding academic performance after enrollment. In this dataset, admission-time factors accounted for 53.2% of the measured lift over the baseline, slightly more than grades and semester progress.
- Python
- pandas
- scikit-learn
- XGBoost
- Jupyter
- seaborn
CSC 487 deep-learning team project
Deep Purring
Can you estimate a cat’s age from its meow?
A playful deep-learning project that turns raw feline vocalizations into log-mel spectrograms and predicts kitten, adult, or senior. My work covered the audio pipeline, baseline models, CNN experiments, and a leakage fix that changed how the whole team split data.
- PyTorch
- librosa
- SmallCNN
- ResNet-18
- SpecAugment
- scikit-learn
EE 428 computer-vision team project
Deepfake Face Detection
Can a detector trained on one kind of fake face recognize another?
The models looked excellent when test images resembled their training data, then became nearly useless on a different deepfake distribution. I built an independent training and visual-analysis pipeline to investigate that generalization collapse in pixel and frequency space.
- PyTorch
- EfficientNet-B3
- ResNet-50
- Xception
- OpenCV
- FFT
Three-person CSC 466 applied-ML project
Potiongram Recommender & Analytics
A fantasy streaming world for exploring recommendations, churn, and user personas.
Across a term-long fictional platform, our team asked three product questions: what should someone watch next, who may cancel, and what behavioral groups exist? I worked on the KNN ablation, the tracked churn pipeline, and the persona outputs.
- Python
- scikit-learn
- pandas
- SVD
- K-Means
- DuckDB
Hack4Impact · Paso Robles Food Co-op
PRFC Referral Program
A referral workflow that helps co-op members invite friends while staff keep track.
Members can send several invitations in one visit while staff review, export, and manage referral activity. My work included the multi-person submission flow, atomic database writes, validation, portal-prefill behavior, and improvements to the staff table and PDF export.
- Next.js 15
- TypeScript
- PostgreSQL
- Prisma
- MUI
- Nodemailer
Six-person software-engineering course project
TaskAdventure
A task manager that turns getting things done into an RPG.
Tasks carry difficulty and stat categories; completing them earns XP and moves a character’s progress. I worked across the MERN stack, modernized the frontend from CRA to Vite, reorganized the backend, tightened auth behavior, and introduced Cypress end-to-end testing.
- React
- Vite
- Express
- MongoDB
- JWT
- Cypress
Four-person database-systems course project
Chore Tracker API
A backend for roommates to coordinate chores and split shared bills.
The API models roommates, chores, assignments, bills, and each person’s share of a payment. I worked on the schema foundation, bill endpoints, assignment flows, synthetic data, concurrency analysis, and a measured indexing investigation at million-row scale.
- FastAPI
- Python
- PostgreSQL
- SQLAlchemy Core
- Docker
- Supabase CLI
- Swagger
